[uf-discuss] citation microformat encodings

[brian suda]

Over the weekend i started to make some serious headway on the citation
microformat. I did run into several properties that
could benefit from some of the microformat schemes we have already
developed. For example, KEYWORDS, SUBJECTS, are both normally a series
of terms that identify the reference. This is akin to rel="tag", so it
makes the most sense to just reuse that microformats. The downside is
that it FORCES the citation to be an 'a href'. Along the same lines as
keywords, there is a property called 'GENRE' this works exactly like
keywords only there is a slightly more structured approach. So does it
make sense to introduce a rel="genre" it works the exact same was as
rel="tag", but genre values are supposed to come from a pre-defined
vocabulary. There is also a 'relation' property which is quite common
through all the different citation formats, it tells how a citation is
related to another. Again this seems an excellent place to use the
rel="" property, similar to XFN rel="friend" a citation could have a
rel="derivative". The downside of using the rel property is that
required a URL back to another source, which may not exist online.

With the rel attribute we can do this,
(there are probably better examples, so don't get hung-up on how i can
reference a dream with a URI)
<a href="http://example.org/doesNotExist/" class="x-relation"
rel="isPartOf">A dream i had last night...</a>

otherwise you'd need to do something like the following:
This is <span class="x-relation"><abbr title="isPartOf"
class="type">part of</abbr> <span class="value">a dream i had last
night...</span>
But this then is not really telling us much, something (class="value")
is a [relationship] that [is part of] that [text-node]... whereas with
the rel attribute, we had started to make a tuple,  [isPartOf] is the
type of [Relationship] to [http://example.org/doesNotExists/]

i have several more issues when it comes to IDENTIFIERS because they can
be of many types, ISBN, ISSN, URI, etc. most of these can be handled,
except in instances of some ambiguity. For Example:

In this case we simply use the node-value of class="x-indentifier"
<!-- we can determine type by the string length --->
ISBN: <span class="x-identifier">1234567890X</span>

In this case it is an 'a' element, so we use the HREF value
<a href="http://example.org/publications/MSc/"
class="x-identifier">http://example.org/publications/MSc/</a>

In this instance it is difficult to determine what is intended, we
shouldn't use the HREF property, it should be the node-value.
<a href="http://amazon.com/books/isbn-here/1234567890X"
class="x-identifier">1234567890X</a>

There are several ways to fix this, one being adding all the different
TYPES of identifiers (url, isbn, issn, etc) i would like to avoid this
because it is an open-ended list of possibilities. The other is to just
NOT allow non-URL identifiers in an 'a' element, or there might be
another idea from this list?

Once i get my notes in order i will post them to the Wiki, in the
mean-time i can field any questions via email off-list.

Thanks,
-brian